Automated Air-Liquid Interface Cell Culture Analysis Using Deep Optical Flow

Autogenerated Report: RAINBOW image series analysis results

Author: Alphons G

Website: https://github.com/AlphonsGwatimba/Automated-Air-Liquid-Interface-Cell-Culture-Analysis-Using-Deep-Optical-Flow

# import required packages 
import json
import os

import cv2

from IPython.display import Image, Video

import pandas as pd

Source directory

print(os.getcwd())

C:\Users\Alphons\OneDrive\Documents\Automated-Air-Liquid-Interface-Cell-Culture-Analysis-Using-Deep-Optical-Flow\examples\image_series_1\(tif) 191018_HNA-ALI_d14.nd2 - 191018_HNA-ALI_d14.nd2 (series 03)0000_etc

Image Series Type

# Load metadata
with open('metadata.json') as f:
    metadata = json.load(f)
    
print(metadata['type'])

.tif

RAINBOW Analysis Timestamp

print(metadata['analysis_timestamp'])

08 April 2021, 12:18:22

Image Series Metadata

print(json.dumps(metadata, sort_keys=True, indent=4))

{
    "analysis_timestamp": "08 April 2021, 12:18:22",
    "calibration_um": 0.31302569743655434,
    "dir": "C:\\Users\\Alphons\\OneDrive\\Documents\\Automated-Air-Liquid-Interface-Cell-Culture-Analysis-Using-Deep-Optical-Flow\\examples\\image_series_1",
    "type": ".tif"
}

Magnitude Statistics

df = pd.read_csv('mag_stats.csv', sep=',')
print(df)

    frame  min (um)   max (um)  mean (um)  std (um)  var (um)
0       0  0.002189   5.659570   2.134026  0.769607  0.592295
1       1  0.007036   6.528444   2.085508  0.828145  0.685824
2       2  0.005678   5.720573   2.050779  0.796154  0.633862
3       3  0.002304   7.910612   1.978159  0.765331  0.585732
4       4  0.001809   6.771280   1.893380  0.784511  0.615457
5       5  0.001108   6.631174   1.770301  0.668799  0.447292
6       6  0.000766   5.486326   1.736731  0.663099  0.439700
7       7  0.007854   7.604922   1.720270  0.686412  0.471162
8       8  0.001544   5.223944   1.766581  0.742917  0.551926
9       9  0.003467   5.256399   1.726751  0.753714  0.568084
10     10  0.005359   7.906752   1.759532  0.706948  0.499775
11     11  0.001898   6.605864   1.763779  0.755759  0.571172
12     12  0.002250   6.330100   1.711699  0.713090  0.508497
13     13  0.001300   8.591791   1.684315  0.720923  0.519731
14     14  0.001470  11.625805   1.684152  0.741490  0.549808
15     15  0.000908   5.909985   1.642674  0.751054  0.564083
16     16  0.002081   6.830723   1.694560  0.724541  0.524959
17     17  0.001529   6.277352   1.709413  0.741429  0.549717
18     18  0.000930   5.813552   1.695623  0.720815  0.519574

Direction Statistics

df = pd.read_csv('dirn_stats.csv', sep=',')
print(df)

    frame  min (deg)   max (deg)  mean (deg)   std (deg)     var (deg)
0       0   0.000291  359.999969  126.017021  124.564008  15516.192018
1       1   0.000028  359.999603  161.746219  135.633649  18396.486804
2       2   0.000493  359.999908  161.909639  131.181341  17208.544188
3       3   0.000007  359.999939  169.133588  133.438504  17805.834264
4       4   0.000105  360.000000  181.327410  130.671843  17075.130625
5       5   0.000625  359.999939  186.161654  121.755672  14824.443715
6       6   0.000034  359.999542  180.747084  117.334445  13767.371877
7       7   0.000129  359.999878  183.374644  119.924125  14381.795800
8       8   0.000388  359.999908  177.643768  123.005852  15130.439508
9       9   0.000008  360.000000  186.818240  118.580904  14061.430882
10     10   0.000012  359.999390  170.768012  118.375267  14012.703734
11     11   0.000349  359.999878  187.630205  118.845236  14124.190120
12     12   0.000058  359.999725  180.344546  117.253174  13748.306783
13     13   0.000041  359.999908  168.804033  114.881794  13197.826541
14     14   0.000123  359.999695  171.097494  117.221801  13740.950567
15     15   0.000390  359.999969  170.326365  113.067118  12784.173157
16     16   0.000809  359.999664  177.686857  117.532094  13813.793091
17     17   0.000206  359.999969  166.113347  114.204230  13042.606063
18     18   0.000655  359.998566  180.108439  112.246061  12599.178306

Raw Image Series (left) and RAINBOW Optical Flow Visualisation (Right)

# video resizing function
def video_reshp(vid_path, set_wdh=None):
    cap = cv2.VideoCapture(vid_path)
    hgt, wdh, _ = cap.read()[1].shape
    dsp_wdh = set_wdh if set_wdh is not None else wdh
    dsp_hgt = dsp_wdh * (hgt / wdh) if wdh is not None else hgt
    
    return dsp_wdh, dsp_hgt

vid_path = 'Combined Images/Combined.mp4v'  
dsp_wdh, dsp_hgt = video_reshp(vid_path, 1280)
Video(vid_path, embed=True, width=dsp_wdh, height=dsp_hgt, html_attributes='controls loop')

a) The direction of motion at any position within RAINBOW generated optical flow images is measured clockwise from the initial horizontal position of a unit circle (left) and is shown using hue values (right).

b) The magnitude of motion at any position within RAINBOW generated optical flow images is shown using saturation values. For instance, high saturation corresponds to high magnitude of motion and low saturation corresponds to low magnitude of motion. For instance, high saturation (100%) corresponds to high magnitude of motion and low saturation (25%) corresponds to low magnitude of motion.

Magnitude Heatmaps Across Image Series

vid_path = 'Heatmaps/heatmap.mp4v'  
dsp_wdh, dsp_hgt = video_reshp(vid_path, 640)
Video(vid_path, embed=True, width=dsp_wdh, height=dsp_hgt, html_attributes='controls loop')

Visualization of motion magnitiude using heatmap with hot colour mapping.

Quiver Plots Across Image Series

vid_path = 'Quiver Plots/quiver_plot.mp4v'  
dsp_wdh, dsp_hgt = video_reshp(vid_path, 640)
Video(vid_path, embed=True, width=dsp_wdh, height=dsp_hgt, html_attributes='controls loop')

Visualisation of optical flow using quiver plots containing vector arrows.